YES! my first job used an in-house object database, and it was magical (though everyone else seemed to hate it.. I think they were just using it wrong.) It had

YES! my first job used an in-house object database, and it was magical (though everyone else seemed to hate it.. I think they were just using it wrong.) It had UML schema, Paxos replication, fine-grained transactions, cardinality checking, reversible upgrades, etc. Application code ran in-process with the database, and it made it really eady to do anything. I'd love to contribute to an open-source version but probably can't since I've read its code. P.S. It fell out of favor because the systems that used it tended to be monoliths that did too much to scale well. The "right way to do it," IMHO is to build a central stateful brain with it, kick all asynchronous work out to stateless worker microservices and read from active secondaries. If the brain's only job is to mutate state by applying business logic, you can scale quite far.